Hu矩和Zernike矩在表情识别应用中的比较  

Comparison of Hu Moment and Zernike Moment in Application of Facial Expression Recognition

作　　者：姜璐[1] 舒华忠[1] 章品正[1] 

机构地区：[1]东南大学生物科学与医学工程系,江苏南京210096

出　　处：《洛阳大学学报》2004年第2期14-17,共4页Journal of Luoyang University

摘　　要：首次将Zernike矩运用于同一个人的表情识别问题,分别计算了同一个人面部图像的Hu矩和Zernike矩,以模式识别中常用的类间距作为依据,提取了面部图像的各种矩的最好特征和次好特征,并对Hu矩和Zernike矩的识别能力和抗噪性进行了比较,试验结果证明Zernike矩更适合识别细微变化的面部表情.By applying Zernike moment in facial expression recognition, Hu moment invariant and Zernike moment invariant respectively in human facial expression images are calculated.According to the criteria of between-to within-class distance, the best and the better feature of several moment invariant in facial expression images and compared the recognition ratio of Zernike moment invariant with Hu moment in variant are extracted. The experimental analyses have proved that the proposed method has the ability to classify facial expression with subtle difference.

关 键 词：表情识别 ZERNIKE矩 HU矩 

分 类 号：O235[理学—运筹学与控制论]